Some Foolishly Precise Predictions About the Bracket

Last year, I ran some numbers on teams from one year and where they return to the next. Some of the work, updated with last year’s tournament, went into my 5 Things You Should Consider When Filling Out Your Bracket post.

I used some aspects of that in order to make some projections about the tournament. It was based on historic trends, patterns, and a three-year moving average. As it turned out, they were pretty good.

I came up with 16 to 18 teams from the ‘07 second round making the ‘08 second round, and 17 ended up doing it.

I projected nine teams from the ‘07 second round making the ‘08 Sweet 16, and 10 did. I projected that from four to six teams (and most likely six) from the ‘07 Sweet 16 would make it that far in ‘08, and five did.

It didn’t work so well when it came to the Elite Eight, since I had fewer than three Elite Eight teams from ‘07 making it back in ‘08, and four teams did. In my defense, the back-to-back years of zero returners in ‘05 and ‘06 messed up the moving average analysis, and it was the first time that the number of repeat Elite Eight teams increased after having been three or more the previous season.

Anyway, I decided to do the same thing again this year. I am also publishing the projections this time just in case it turns out to be good again. I also applied the same analysis to seed in the first round, so we’ll see how this goes.

It could be that this is snake oil and I got lucky last year, or it could be really close again. We’ll have to see. Anyway, here’s what I got for this year:

• 15 teams from the ‘08 second round in the ‘09 second round
• From seven to nine (likely eight) teams from the ‘08 second round in the ‘09 Sweet 16
• Five to seven (likely six or seven) teams from the ‘08 Sweet 16 in the ‘09 Sweet 16
• Between one and three (likely one) teams from the ‘08 Elite Eight in the ‘09 Elite Eight
• One team from the ‘08 Elite Eight in the ‘09 Final Four
• Either five or six (likely six) upsets in the ‘09 first round

If there are five first round upsets, the second round will have four ones, four twos, four threes, three fours, three fives, three sixes, two sevens, one eight, three nines, two tens, zero elevens, three twelves, one thirteen, and zero each of fourteens, fifteens, and sixteens.

If there are six first round upsets, the second round will have the same as above only with one five, four sixes, three twelves, and zero elevens.

As the post title indicates, I have a feeling that this is probably a fool’s errand. However, I’m hoping when the brackets come out that the combination of the returners and seeds will put together a pretty accurate projection of the field. I think it’s probably more likely that it will be impossible to meet all of the above conditions.

After all, if there’s one thing that I know about the tournament, it changes everything up the moment you think you have it figured out.

Five Things To Consider When Filling Out Your Bracket

It’s almost bracket picking time, and everyone has a system or a theory to help them out. Regardless of your system, consider these historical trends while you’re picking your winners.

Keep in mind that these trends say what is likely to happen, not what will happen. Everything here is based off of the era of 64-team tournaments, which means we’ve only got 24 past tournaments to go off of.

One last thing: I do not classify an eight-seed losing to a nine-seed as an upset. That is all; let’s do this.

Kansas, UCLA, Memphis, and North Carolina will all win their first round games, but at least one will probably lose its second round game.

In the last ten years, nearly every Final Four team has won its first round game the next year (provided it made the tournament). The three that have not were all Big Ten teams and six-seeds or below. No Big Ten squads made the Final Four and it’s looking like all of last year’s participants will be at least four-seeds, so those teams should be safe.

For all of the tournaments though, never have all Final Four teams from one year made the Sweet 16 the next.

The champion will almost certainly be a one, two, or three-seed.

Only three teams lower than a 3-seed have won it all: 8-seed Villanova in 1985, 6-seed Kansas in 1988, and 4-seed Arizona in 1997. Keep in mind that in the ‘80s when the six and eight-seeds won, we didn’t have nearly the coverage of the sport we do now. The committee has gotten better with more time and more film, and a team at the top will take home the title.

In case you’re wondering, one-seeds have won just over half of the championships and seven of the last ten. Three-seeds aren’t even that great a bet, as only three of those have ever won the whole thing (though two were this decade).

Strictly speaking, based on history each one-seed has a 13.5 percent chance of winning it all, each two-seed has a 5.2 percent chance, each three-seed has a 3.1 percent chance, and everyone else from four-seeds to eight-seeds has a 0.4 percent chance.

One and three-seeds playing in their home state are money, but twos are not quite so reliable.

Only a single one-seed in 62 contests has lost a game in its home state, and that was in 2001 when three-seed Maryland beat Stanford in Anaheim. Only a single three-seed in 19 contests has lost a game in its home state, and that was in 2007 when Texas A&M lost to two-seed Memphis in San Antonio.

Two-seeds however are just 30-8 (.789) in their home states, essentially losing one of every five contests. When you take out games against one-seeds, they go to 27-7 (.794) which is basically the same performance. They have won ten in a row in their home state, but six of those came from UCLA’s 2006 and 2007 teams. Even then, two-seeds are just 16-4 (.800) in home state games since 2003.

When picking first round upsets, don’t bet on lucky seven.

The average number of first round upsets is 5.63. The most common number is five in a year, something that has happened seven times. The next most common number of first round upsets is eight (five times) and then six (four times).

What about seven you ask? We’ve seen seven first round upsets exactly once, and that was in 2002. I have no good explanation for this phenomenon other than that there have been a relatively small number of 64-team tournaments, but try not to bet against history with this one.

For what it’s worth, this decade has evenly split up the number of upsets: 2007 had two, 2004 had three, 2005 had four, 2003 had five, 2008 had six, 2002 had seven, 2006 had eight, and 2001 had nine. If you’re considering extending the pattern, be advised that we’ve never seen one or ten first round upsets in a year. Then again, we had never seen two or nine in a year until it happened this decade.

Having a team return to the Final Four is about a coin flip.

Having all Final Four teams shut out of the next year’s Final Four has happened 11 times in 23 possible chances. That means 12 times in 23 chances we’ve seen at least one come back.

The most that have ever returned is two, and each time that has happened one of the two Final Four repeat teams was on at least a three-year run of making it that far: Duke and UNLV both made it in 1990-91 during a four-year run for Duke, Kentucky and North Carolina both made it in 1997-98 during a three-year run for UK, and both UCLA and Florida made it in 2006-07 during a three-year run for the Bruins.

That would seem to indicate that if two teams were to make it back, UCLA would be one of them. It’s a trend, not a rule though, so nothing is set in stone.

Anyway, it’s no guarantee than any of last year’s bunch makes it back, much less two. Just pull out a quarter and let George tell you.

Points per Drive in 2008

Perhaps the most important thing you can do in football is to maximize the return of your offensive possessions. You only get so many per game, and you don’t fully control how many you get. If your opponent is determined to sit on the ball for most of the contest, you simply won’t get as many chances to score as you otherwise would.

Some people may disagree with that though. They may argue that the most important thing you can do in football is to ensure your opponent gets the least out of their possessions as possible. A stifling defense can make up for offensive struggles and give the offense more possessions with which to work.

Regardless of which side you believe in, the same stat can be used to figure out how well your team is doing at both: points per drive. It’s not perfect since things like special teams and turnovers can affect that stat, but I think I can show that it’s pretty darn good at measuring how good a team is.

To calculate points per drive, you need two parts: points and the number of drives. Figuring out points is the easy part since you just look at field goals, rushing touchdowns, and passing touchdowns. That filters out special teams and defensive touchdowns.

I left out extra points and two point conversions because they have little to do with how offenses and defenses truly perform over the course of a game. I had no choice but to leave in lost/gained fumbles in special teams situations since there are no stat sources that separate them out. I’m mostly fine with that though since gaining or losing a fumble in special teams results in gaining or losing a possession. PATs on the other hand do not have anything to do with possession counts.

To calculate number of drives, I added up the following categories: punts, fumbles lost, interceptions, failed fourth down conversions, field goal attempts, and touchdowns of the rushing and passing variety. The NCAA official stats only have the offensive version of these stats and a few of the defensive, but the fantastic site cfbstats.com fills in the rest.

The top ten teams in offensive points per drive were the following:

1. Texas Tech – 3.27 points per drive
2. Oklahoma – 3.24
3. Florida – 3.22
4. Texas – 3.17
5. Tulsa – 3.12
6. Oklahoma State – 2.92
7. Missouri – 2.89
8. Penn State – 2.80
9. Rice – 2.70
10. Ball State – 2.698

fansnap_syndslot_get_headliner();
fansnap_syndslot_get_category();
var fansnap_syndslot_instance = new fansnap_syndslot();
var fansnap_script_include = new Element(’script’);
fansnap_script_include.src = fansnap_syndslot_instance.server + ‘/synd/js/search?’ + fansnap_syndslot_instance.params();
$(‘fansnap-instance’).insert(fansnap_script_include);

If you read any of my pieces on pace, this list will look familiar. All of these teams were either in the top ten of yards per play or points per play. The fact that they appear here should not be a surprise.

Here are the top ten teams in defensive points per drive:

1. USC – 0.65 points per drive allowed
2. Boise State – 0.75
3. TCU – 0.79
4. Iowa – 0.95
5. Alabama – 0.98
6. Ohio State – 0.99
7. Florida – 1.00
8. Boston College – 1.02
9. Penn State – 1.09
10. Utah – 1.13

Interestingly enough, Florida and Penn State are the only two teams in the top ten of both. If I said earlier that there were two teams were in both, I’d imagine many people would pick out UF, but probably not Penn State.

In any event, these two measures are not infallible predictors of great success. Houston (8-5) was 11th and Arizona (8-5) was 14th in offensive points per drive. Tennessee (5-7) was 11th and Clemson (7-6) was 13th in defensive points per drive.

In order to find out how good of predictors of success these measures are, I decided to run correlations for them with winning percentage.

I fully expected to see the negative correlation of defensive points per drive to be stronger than the positive correlation of offensive points per drive with win percentage. After all, how many times have we all heard that defense wins championships? Probably more than we can count.

Having been a believer in that myself, what I found shocked me.

The correlation of offensive points per drive and win percentage was 0.715. The correlation of defensive points per drive and win percentage was -0.711. In other words, there basically is no difference in their ability to predict success. Offense and defense are equally important.

Now, that is only one year’s worth of data at work. I can really only say that offense and defense were equally important in 2008. I am in the process of running data on past years to try to get a better idea of how they relate. For now though, they’re equal.

The next obvious step was to try to synthesize these two into one measure to see how high a correlation I could get. It is not as simple as adding the two numbers together because with one, a high number is good and with the other, having a low number is good.

I chose to go the route of deviation from the mean. So for offensive points per drive, I simply subtracted the mean from the team’s number. For defensive points per drive, I subtracted the team’s number from the mean. I then added those two together.

That process gave a single number to correlate with winning percentage. For now I’m calling it combined points per drive, but if you have a better name, leave it in the comments.

First, here are the top ten teams in combined points per drive:

1. Florida – 2.12
2. USC – 1.90
3. Texas – 1.68
4. Oklahoma – 1.65
5. Boise State – 1.63
6. Penn State – 1.61
7. TCU – 1.46
8. Tulsa – 1.25
9. Texas Tech – 1.222
10. Utah – 1.220

The worst record among them is Tulsa’s 11-3 mark. All eleven teams with a winning percentage above .800 are contained in the top 13 spots.

Given those observations, it should come as no surprise that the correlation between combined points per drive and win percentage is 0.923 for 2008. That is an extremely high correlation and about as high as you can expect for just two stats put together.

The important thing to remember is that this describes what teams did in the context of their opponents. This stat has not been adjusted for strength of schedule, so it would not make sense to take the above list and proclaim that Boise State is better than Penn State because of a 0.02 difference between them. A WAC schedule just doesn’t compare favorably to a Big Ten schedule, regardless of what you think of the current state of the Big Ten.

The top four beg some sort of interpretation though. It puts Florida and USC well ahead of the top two Big 12 teams, Texas and Oklahoma.

One way to interpret it is to say that Florida and USC were the two best teams and should have played for the national title. After all, there really isn’t that big a difference between the top five or six conferences.

Another way to interpret it is to point out that the Pac-10 and SEC were down, while the Big 12 had perhaps the best year of its entire existence. Of course Texas and Oklahoma would be lower; they played in the season’s toughest conference.

As I said I haven’t adjusted for schedule strength, so until and unless that happens, the debate will remain open. For what it’s worth, the NCAA says Oklahoma, Florida, and Texas (in that order) had the top three toughest schedules. USC’s slate clocked in at No. 38.

Regardless, the stat of combined points per drive seems to be a very accurate indicator of what degree of a winner a team was. I plan to explore this one further to see what else it might hold in store.

A Wrapup On Pace in 2008

The past four days I’ve posted topics on how pace affected football in 2008. The primary impetus for doing the series was to put Oklahoma and Tulsa in their proper historical context.

Oklahoma scored the most points ever in a season, and Tulsa gained the most total yards ever in a season. When two records like that fall in the same season, especially one where a clock rule change reduced plays per game and scoring from the old rules, it’s worth taking a look to see why that might have happened.

The easiest answer is that both OU and Tulsa played in 14 games. The twelfth game added to the schedule earlier this decade, when combined with conference championship games and bowls stats counting towards season totals, basically meant that it was a matter of time before some of these records fell. Anything set back when the season had only 11 games and bowl stats didn’t count towards season stats was doomed.

The extra game doesn’t tell the whole story, though. Each of these records, both points scored and yards gained, were set by 2006 Hawai’i. That team played 14 games, and that season’s clock rules lowered plays per game and scoring even more than 2008’s clock rules did. The extra game helped OU and Tulsa pass most teams, but it was not the deciding factor in breaking the records.

That is where playing at a faster pace comes in.

There are distinct advantages to running a hurry-up offense full-time beyond just getting more opportunities to score. When you go at a faster pace, you can disrupt the defense and gain an advantage. The defense may not be set every time and it will not be able to substitute as often. Plus, your team is better conditioned to play at the faster pace than a team that doesn’t, so you can tire out the other side too.

Whether each team has six drives apiece or 15 drives apiece during a game, you still want to score on more of them than the other guy. Cranking up the pace is done with the idea of gaining an advantage that you cannot get at a normal pace and exploiting it to score more often than the other team.

Oklahoma chose to turn up the pace in response to the new 40 second play clock. Gus Malzahn of Tulsa has long been a proponent of the hurry-up, and you can purchase his book on the topic on Amazon. The end result of each team’s fast paced attacks was two big records falling.

As I mentioned yesterday, 1989 Houston still holds the record for points per game. That 2006 Hawai’i team that used to hold the total yards and points records still holds the yards per play record at 8.6 as well. Since Tulsa and Oklahoma do not now hold the rate records, only the total records, it is reasonable to conclude that the records fell almost entirely because of each team’s fast pace allowing them to run more plays than teams in the past.

I want to be clear about one thing though. I am not trying to bring down either of these teams. Each turned in remarkable offensive seasons that are among the greatest college football has ever seen.

There also is no way of knowing if those Houston and Hawai’i teams of the past could have kept up their rates at the faster pace either. After all, 1970 Notre Dame holds the plays per game record at an astonishing 92.4, but that Irish team doesn’t hold any other records to go with it. It’s one thing to theorize what a team could do, but it’s another to actually do it.

Bill Simmons of ESPN.com wrote a piece recently on the way that Mike D’Antoni’s “seven seconds or less” offense affected stats in the NBA. The most dramatic effect was taking Steve Nash and turning him from a good point guard into one of only nine players ever to win back-to-back MVPs. Simmons then showed that Nash’s stats from this year without D’Antoni are nearly identical to his stats from his pre-D’Antoni Dallas days.

As a fan, I have absolutely nothing against fast-paced offenses. I loved watching D’Antoni’s Phoenix Suns teams, and what I got to see from Oklahoma and Tulsa this year was very exciting as well. Kevin Wilson and Gus Malzahn appeared to maximize the talent they had with their uptempo schemes, and that’s a beautiful thing to see any time it happens.

At the same time, it’s important to realize the distinction between the NBA and college football. What people think doesn’t matter in the team sense in the NBA thanks to the league having a playoff. College football determines its champion largely thanks to opinion polls, so what people think does matter.

I can’t think of a year in which what people thought mattered more than it did with Oklahoma and Texas this year. I don’t mean to rehash old news, but Oklahoma’s impressive scores were largely the reason why it passed up the Texas team that beat it earlier in the year. That then allowed Oklahoma to go to the Big 12 title game and on to the BCS title game.

It’s possible that had OU operated at a slower pace and didn’t put up 60 points in five straight games, it might not have passed up UT. If that doesn’t happen, Texas likely beats Mizzou in the Big 12 title game and goes on to play Florida for the national title.

There’s no way to know, but Texas could have beaten Florida and won the national title. If Texas makes the national title game, then Colt McCoy probably wins the Heisman trophy as well. So, not only did pace potentially affect the Heisman race as it affected the NBA’s MVP race, but it potentially affected the championship.

It’s not likely we’ll ever get all of the voters to look at efficiency stats like points per play or points per drive instead of final scores, so as long as the BCS exists, this same thing can happen again. The moral of the story is that cranking up the pace is a fantastic way to game the system if you can pull it off, and for the record, I’m all for gaming the system.

The 2008 Oklahoma and Tulsa offenses are the two most prolific we’ve ever seen at generating points and yards. They were special, and no one can deny that. They were not uniquely special in the annals of the game though, and that’s the takeaway for thinking about the ‘08 season in historical context.

2008 Scoring at Oklahoma’s Pace

Pace was one of the hot button issues in the 2008 college football season. Oklahoma’s highly publicized switch to a fast paced offense in reaction to the new clock rules was the major reason for it. The Sooners ended up leading the country in plays at 1,106 (79 per game), and they set a record with 716 total points scored.

The Sooners weren’t the only team to crank it up. Tulsa, under no-huddle guru Gus Malzahn, was second in plays behind OU, and Houston, TCU, and Nevada also broke 1,000 plays for the season.

The average number of plays per team for the whole season was 858.52. The average number of games played was 12.68. Therefore, the average number of plays per game for any given team was 67.7.

But what if everyone played at Oklahoma’s pace? Here is a look at what the top ten in scoring would look like if everyone ran 79 offensive plays a game.

This would be the point where I mention that this is based off of the NCAA’s “scoring offense” stat, which includes defense and special teams scores in the totals. Because this study is looking at pace in terms of plays, and it proportionately increases or decreases each team’s total plays, it still works out under the assumption that teams would continue to get defense or special teams scores at the same pace as before.

The top ten in scoring, adjusted to be at Oklahoma’s 2008 pace:

Top Ten Points per Game at Oklahoma’s Pace

Team	Total Pts	Pts/Game	Pts/Play	Adj. Total Pts	Adj. Pts/Game
Florida	611	43.64	0.70	773	55.23
Oklahoma	716	51.14	0.65	716	51.14
Tulsa	661	47.21	0.60	666	47.60
Missouri	591	42.21	0.60	666	47.54
Oklahoma St.	530	40.77	0.58	599	46.11
Texas Tech	569	43.77	0.58	597	45.92
Texas	551	42.38	0.58	593	45.58
Oregon	545	41.92	0.57	584	44.90
Penn St.	506	38.92	0.57	581	44.71
Rice	537	41.31	0.56	572	43.96

Tulsa edges out Missouri in points per game, even though rounding to the nearest point makes them equals in total scored.

What we can see here is that Oklahoma was ahead of pretty much everyone at scoring points. Adjusting for pace, they still were ahead of most of the nation and earned their record 716 points scored.

Florida was the one exception. Thanks to getting points in many ways other than just offense (INT returns, off of blocked punts, in the return game, etc.) while running about an average number of offensive plays, Florida would have shattered the Sooners’ new record in the very year they broke it.

The Gators would have topped out at a little over 55 points a game. That means Army’s all-time record would have been safe, but only barely. In 1944, Army scored exactly 56 a game, less than a point than Florida’s hypothetical total.

It is almost a little surprising to see Missouri so high since the Tigers were a bit of a disappointment this season. It goes to show that the offense was still good at turning plays into scores, but that defense just didn’t quite work out.

As great as Florida and Oklahoma were at turning plays into points by having relatively high points per play ratios, they weren’t the best of the decade. Since 2000, the team with the highest PPP was 2006 Hawai’i, with 0.72 points per play. At Oklahoma’s pace over 14 games, that would come out to 795 points on the season.

One would figure though that if they were that close to 800, they’d find a way to get one last touchdown to get to 802. Maybe something like the Florida Flop?

2008 Yardage at Oklahoma’s Pace

Pace was one of the hot button issues in the 2008 college football season. Oklahoma’s highly publicized switch to a fast paced offense in reaction to the new clock rules was the major reason for it. The Sooners ended up leading the country in plays at 1,106 (79 per game), and they set a record with 716 total points scored.

The Sooners weren’t the only team to crank it up. Tulsa, under no-huddle guru Gus Malzahn, was second in plays behind OU, and Houston, TCU, and Nevada also broke 1,000 plays for the season.

The average number of plays per team for the whole season was 858.52. The average number of games played was 12.68. Therefore, the average number of plays per game for any given team was 67.7.

But what if everyone played at Oklahoma’s pace? Here is a look at what the top ten yardage gainers would look like if everyone ran 79 offensive plays a game.

Top Ten Yards per Game at Oklahoma’s Pace

Team	Total Yds	Yds/Game	Yds/Play	Adj. Total Yds	Adj. Yds/Game
Tulsa	7,978	569.86	7.27	8,043	574.53
Houston	7,316	562.77	7.20	7,395	568.86
Florida	6,231	445.07	7.13	7,885	563.21
Texas Tech	6,903	531.00	7.05	7,241	557.03
Oklahoma St.	6,340	487.69	6.98	7,171	551.61
Oklahoma	7,760	547.86	6.93	7,670	547.86
Missouri	6,778	484.14	6.90	7,634	545.28
Georgia	5,538	426.00	6.70	6,886	529.66
Ball St.	6,195	442.50	6.70	7,407	529.09
USC	5,911	454.69	6.63	6,813	524.10

The Tulsa Golden Hurricane tops the list at an incredible 8,043 yards for the season. The all-time record, if you’re wondering, was 7,826 set by 2006 Hawai’i before Tulsa broke it with its actual 7,978 yards in 2008. However, the per-game record of 624.9 set by 1989 Houston is still safe in theory as well in actuality.

The appearance of two SEC teams on this list while not appearing on the actual list shows that run-first, slower paced conferences can still produce some efficient offenses. That fact was lost on a lot of people when picking the national title game, as many saw Oklahoma as clearly the better offensive team. The Sooners were definitely more prolific, but we can see here that the Gators were more efficient.

Everyone on this list averaged more than 524 yards a game at Oklahoma’s pace. In real life, only four teams averaged that much: Tulsa, Houston, Texas Tech, and OU. Only one other team, Nevada, averaged more than 500 real yards a game.

The presence of Georgia, Ball State, and USC also show that pro-style offenses can be highly efficient just like the spread offenses that are all the rage. You likely won’t hit Gus Malzahn-like pinball numbers, but there is something to be said for doing it the old fashioned way. It still gets the job done.

I don’t know if we can really learn much from this, but it’s still fun to look at and think about how close Tulsa was to getting to eight grand. Malzahn may have left Tulsa for Auburn, but Oklahoma returns a lot of tools from its team last season. What do you say, Bob Stoops and Kevin Wilson? Why not make a run at 8,000 yards next year?

2008 Scoring Adjusted for Pace

Pace was one of the hot button issues in the 2008 college football season. Oklahoma’s highly publicized switch to a fast paced offense in reaction to the new clock rules was the major reason for it. The Sooners ended up leading the country in plays at 1,106 (79 per game), and they set a record with 716 total points scored.

The Sooners weren’t the only team to crank it up. Tulsa, under no-huddle guru Gus Malzahn, was second in plays behind OU, and Houston, TCU, and Nevada also broke 1,000 plays for the season.

The average number of plays per team for the whole season was 858.52. The average number of games played was 12.68. Therefore, the average number of plays per game for any given team was 67.7.

To take a look at how well everyone was able to score points on equal footing, I have adjusted total points by pace.

This would be the point where I mention that this is based off of the NCAA’s “scoring offense” stat, which includes defense and special teams scores in the totals. Because this study is looking at pace in terms of plays, and it proportionately increases or decreases each team’s total plays, it still works out under the assumption that teams would continue to get defense or special teams scores at the same pace as before.

Here is a table showing the top ten teams in points per game if everyone played at the nation’s average pace in terms of plays per game.

Top Ten Adj. Points per Game

Team	Total Pts	Pts/Game	Pts/Play	Adj. Total Pts	Adj. Pts/Game
Florida	611	43.64	0.70	663	47.33
Oklahoma	716	51.14	0.65	613	43.83
Tulsa	661	47.21	0.60	571	40.79
Missouri	591	42.21	0.60	570	40.75
Oklahoma St.	530	40.77	0.58	514	39.52
Texas Tech	569	43.77	0.58	512	39.35
Texas	551	42.38	0.58	508	39.06
Oregon	545	41.92	0.57	500	38.48
Penn St.	506	38.92	0.57	498	38.32
Rice	537	41.31	0.56	490	37.68

Here we can see that when pace is accounted for, Florida usurps Oklahoma as the total points leader as OU loses over a hundred points. The accelerated pace that the Sooner offense operated at allowed them to score about a touchdown a game more than they would have had they played at the average pace. On a per game basis, the Gators were first by a little more than a field goal over OU.

The Big 12 is well represented with five of the top ten. Oregon is the only Pac-10 team in the top ten, but USC is lurking just outside at No. 12.

Penn State meanwhile is the only Big Ten team here, and the next-highest conference colleague of the Nittany Lions’ was Iowa at No. 27. In numerical terms, the Hawkeyes are almost a full touchdown behind. That shows just how much more efficient PSU was than the rest of its conference.

Florida holds a similar status in the SEC since it was clearly the most efficient team at putting points on the scoreboard. Florida was two tenths of a point away from being a full two touchdowns ahead of the next highest SEC team, No. 21 Georgia.

The ACC and Big East are the only BCS conferences not represented here. FSU was the highest ACC team at 20th, but their numbers are skewed a bit by 115 points scored in its first two games against I-AA opponents. Those 115 points make up 26.5% of the Seminoles’ total points in 2008. The Big East’s highest team was actually Rutgers, at 33rd overall.

I can’t finish this without a word about Rice. It’s amazing what Chase Clement and Jarrett Dillard were able to do for that team over their careers there. “Rice football” has been synonymous with “losing” in Texas for many years, but the Owls won 10 games a year ago. Rice has some other players, but those two will be sorely missed next season.

2008 Yardage Adjusted for Pace

Pace was one of the hot button issues in the 2008 college football season. Oklahoma’s highly publicized switch to a fast paced offense in reaction to the new clock rules was the major reason for it. The Sooners ended up leading the country in plays at 1,106 (79 per game), and they set a record with 716 total points scored.

The Sooners weren’t the only team to crank it up. Tulsa, under no-huddle guru Gus Malzahn, was second in plays behind OU, and Houston, TCU, and Nevada also broke 1,000 plays for the season.

The average number of plays per team for the whole season was 858.52. The average number of games played was 12.68. Therefore, the average number of plays per game for any given team was 67.7.

To take a look at how well everyone was able to gain yards on equal footing, I have adjusted total yards by pace. Here is a table showing the top ten teams in yards per game if everyone played at the nation’s average pace in terms of plays per game.

Top Ten Adj. Yards per Game

Team	Total Yds	Yds/Game	Yds/Play	Adj. Total Yds	Adj. Yds/Game
Tulsa	7,978	569.86	7.27	6,893	492.38
Houston	7,316	562.77	7.20	6,338	487.52
Florida	6,231	445.07	7.13	6,757	482.68
Texas Tech	6,903	531.00	7.05	6,205	477.38
Oklahoma St.	6,340	487.69	6.98	6,145	472.73
Oklahoma	7,760	547.86	6.93	6,573	469.51
Missouri	6,778	484.14	6.90	6,542	467.30
Georgia	5,538	426.00	6.70	5,901	453.92
Ball St.	6,195	442.50	6.70	6,348	453.43
USC	5,911	454.69	6.63	5,839	449.15

As you can see, no one was able to rack up yards better than the guy who wrote the book on the hurry up offense. In Houston it was no Art Briles, no problem for the Cougars under first year head coach Kevin Sumlin. We can also see that Florida played under the national pace in 2008, which is likely a side effect of the more defense- and ball control-oriented SEC.

A run of four Big 12 teams with nearly identical marks come next before we get to another SEC team in Georgia. The Bulldogs were also below the national average pace, and Ball State was the only other squad on the list below that average. It shows that these teams had the capacity to rack up yardage comparable to the Big 12 and CUSA teams on the list, but either chose not to for philosophical reasons or were prevented from it by their opponents’ pace (or both).

USC surprised me at first not for being in the top ten, since Mark Sanchez was great and the Pac-10 was down, but for being above the national pace. The Trojans always field good offenses, but they don’t stick out to me as being in a hurry when they play.

That’s when it occurred to me. USC had a historically good defense this season, ranking second only to the 2001 Miami (FL) defense in terms of yards per point allowed in this decade. The fact that the Trojans’ defense caused a lot of three-and-outs probably pushed their plays per game above average. It goes to show that plays per game has to do with more than just offensive pace (though for these purposes, it’s close enough).

Of course, football is not won and lost on yards gained. I also adjusted points scored for pace as well. If everyone played at the national average pace, would Oklahoma still come out on top in total points scored?

Stay tuned.

Converting Between the College, NFL Passer Ratings

As long as quarterbacks have played the central role of offensive football, people have tried to quantify who is the best. Various methods have been concocted to do just that, and many more are being devised even today.

The two most widely-cited measures are passer rating and passing efficiency. The former is used by the NFL, while the latter is used by the NCAA.

They both are complex formulas, and if you want the details, hit up the passer rating Wikipedia page. Despite their differences, they use the same four components: completion percentage, yards per attempt, touchdowns per attempt, and interceptions per attempt. What differs is how the parts are weighted.

The NFL’s passer rating imposes a ceiling and floor on the four parts, so it has a lower boundary at zero and an upper boundary of 158.3. The idea is not to let outliers, good or bad, have undue influence on the rating. If you’re curious, the answer is yes, the pro game has seen its share of both perfect games and zero games.

The NCAA’s passing efficiency has no such boundaries against outliers. The maximum score occurs when someone completes every pass for a 99-yard touchdown and the minimum score occurs when someone completes every pass for a 99-yard loss. You are correct in assuming we’ve never seen anyone log a maximum or minimum score.

Despite there being pros and cons to each system, they are generally kept apart. The passer ratings of college quarterbacks and the passing efficiency of NFL quarterbacks are not widely reported.

Here are some tables that show some insight into how the systems differ and how we might compare the relative performances of collegiate and professional quarterbacks.

For the sake of brevity, I have included only the top ten of each category in the tables.

I Said Relative Performance

Before we get into numbers, I want to stress that any comparisons done between college and pro quarterbacks are meant to viewed in relative terms.

The NFL obviously has tougher defenses than college does, but the NFL also has better offensive lines and, well, quarterbacks too. I don’t think anyone would argue that the Peyton Manning of today is not better than the version of himself that lost to Florida four times at Tennessee.

Take the inter-division comparisons with a grain of salt, and know that this (like football) is in the end just for fun.

NFL Passing Efficiencies

I will start with passing efficiency of the primary NFL starting quarterbacks. I got my stats on them from ESPN’s stats page for the regular season, so if you’re looking for the passer rating standings, there you go.

2008 NFL Passing Efficiency

Rank	Player	Team	Passing Efficiency	Pass. Rat. Rank
1	P. Rivers	SD	154.6	1
2	D. Brees	NO	144.4	4
3	K. Warner	ARI	143.3	3
4	C. Pennington	MIA	142.1	2
5	M. Schaub	HOU	141.1	7
6	A. Rodgers	GB	139.3	6
7	P. Manning	IND	139.1	5
8	T. Romo	DAL	138.5	8
9	M. Ryan	ATL	134.7	11
10	J. Garcia	TB	132.8	9

Here, Philip Rivers still rules the roost. There’s a little movement in the rankings, but no one slides more than two spots one way or the other.

None of these numbers really pop out though, even Rivers’ mark. That is because college quarterbacks routinely achieve loftier numbers, such as Sam Bradford’s 180.3 mark that led the college game in 2008.

For comparison, Rivers’ efficiency score would land him at 14th-best in the country between Ball State’s Nate Davis and Nebraska’s Joe Ganz. There is a good reason why college quarterbacks can go higher than the pro guys, and while I think you know what it is, I’ll take a look at it later.

The lowest passer efficiency score was by Cleveland’s Derek Anderson. He managed a 103.0 passing efficiency. By comparison, the 100th-ranked college passer was Kentucky’s Mike Hartline with a 104.7 score.

College Passer Ratings

Now it’s time to see how the big men on campus fared using the NFL’s report card. Their stats came from the NCAA stats site.

2008 NCAA I-A Passer Ratings

Rank	Player	Team	Passer Rating	Pass. Eff. Rank
1	S. Bradford	OU	127.0	1
2	T. Tebow	UF	122.1	4
3	C. McCoy	TEXAS	121.6	3
4	D. Johnson	TULSA	117.7	2
5	C. Clement	RICE	116.5	7
6	M. Sanchez	USC	113.0	6
7	G. Harrell	TTU	112.9	8
8	C. Keenum	HOU	110.9	9
9	Z. Robinson	OKST	110.2	5
10	C. Daniel	MIZZ	107.5	10

As you can see, the college guys do better overall on the NFL’s scale too. In fact, Bradford’s season would shatter Peyton Manning’s all-time record of 121.1 for a single season. The other two Heisman finalists would edge him out too, for that matter.

There was a bit more movement in these standings after conversion than in the NFL standings, with Oklahoma State’s Zac Robinson taking the biggest fall at four spots. I don’t know if that has more to do with formulaic differences, but I have a feeling it has more to do with the fact that there are a lot of quarterbacks in I-A college football. The bunching that ensues means small real drops could get magnified as relative drops.

The lowest passer rating in the pros was by the Browns’ Anderson again with a score of 66.5. Kentucky’s Hartline, Mr. 100th Place in college, had a rating of 69.4.

You know there has to be something inflating the college stats. I mean the No. 32 college quarterback in passer rating was Illinois’ Juice Williams, and he managed to post an 86.4 rating. That would tie him for 14th place in the NFL with Eli Manning and Donovan McNabb.

Adjusted College Passer Ratings

The inflation factor was something we all know. They’re sweet, they’re fluffy, they’re cupcakes.

As I said above, I’m looking to judge relative value. NFL teams don’t get to stock up to a third of their schedule with arena league teams, but the top college teams can schedule anywhere from three to five teams (depending on the conference) that cannot compete on the top team’s level.

In 2008, the power conferences were the six BCS leagues plus the Mountain West Conference. Because I’m feeling charitable, and because their name is in the BCS contracts too, I counted the Fighting Irish of Notre Dame as a power team for this part too.

So, I took the top ten quarterbacks from these conferences and took out all stats against teams that are not power teams. You can argue that in 2008 Central Michigan and Troy were much better foes than, say, Washington or Washington State, and you’d probably be right. Even so, I had to draw the line somewhere.

Here is what the passing efficiency stats look like for the top college quarterbacks from power teams against power teams:

Top QBs from Power Teams Against Power Teams

Rank	Player	Team	Pass. Eff. Adj.	Pass. Eff.	Diff.
1	S. Bradford	OU	181.0	180.8	+0.2
2	T. Tebow	UF	166.3	172.4	-6.0
3	M. Sanchez	USC	164.6	164.6	-
4	C. McCoy	TEXAS	163.0	173.8	-10.8
5	G. Harrell	TTU	162.7	160.0	2.7
6	Z. Robinson	OKST	155.8	166.8	-11.0
7	J. Ganz	NEB	153.4	153.7	-0.3
8	M. Stafford	UGA	150.6	153.5	-3.0
9	B. Johnson	UTAH	148.4	149.4	-1.0
10	C. Daniel	MIZZ	146.7	159.4	-12.7

Bradford’s and Sanchez’s numbers didn’t change much because Oklahoma played only one cupcake (I-AA Chattanooga) and USC didn’t play any.

You can see, however, that three of the other five Big 12 quarterbacks and Tebow benefited some from feasting on weaker, non-conference competition. At least in Tebow’s case he didn’t fall behind anyone as a result. No one else changed that dramatically, though Texas Tech’s Graham Harrell somehow got better against better competition.

Philip Rivers moves up into seventh place now that we’ve focused on quarterbacks from the top of Div. I-A and only how they do against other top teams. The college quarterbacks still have crappy BCS conference teams on their side, but at least the empty calories have been removed.

Finally, let’s take a look at the passer ratings of the college players.

Adjusted College Passer Ratings

This table contains the same guys, only this time it’s using the NFL’s system.

Top QBs from Power Teams Against Power Teams

Rank	Player	Team	Pass. Rat. Adj.	Pass. Rat.	Diff.
1	S. Bradford	OU	126.8	127.0	1
2	T. Tebow	UF	117.6	122.1	-4.5
3	G. Harrell	TTU	115.9	112.9	3.0
4	M. Sanchez	USC	113.0	113.0	-
5	C. McCoy	TEXAS	112.0	121.6	-9.6
6	J. Ganz	NEB	103.8	103.0	0.9
7	B. Johnson	UTAH	103.2	103.5	-0.3
8	Z. Robinson	OKST	103.0	110.2	-7.2
9	M. Stafford	UGA	98.5	101.7	-3.2
10	C. Daniel	MIZZ	97.2	107.5	-10.3

So Peyton’s record is still falling at the hands of the new Heisman winner, but no one else is breaking it this year. Philip Rivers also moves up a spot to sixth, behind only the three Heisman finalists, USC’s new blue chipper, and a guy who runs an offense called the “Air Raid.” Not bad, Phil.

Missouri’s Chase Daniel again takes a hard hit in the rankings. This is no surprise to readers of the excellent Dr. Saturday site, where editor Matt Hinton showed that Daniel was only able to light up bad defenses this past season. Maybe it was the thumb injury, or maybe he wasn’t that good. I don’t know if we’ll ever find out.

Even as Daniel struggled to post big numbers against teams with a pulse, his adjusted passer rating was still higher than 30 of the 32 regular starting NFL quarterbacks. Why are there so many college quarterbacks with monster passer ratings?

Think Spectrums

I don’t mean to keep singling out Mike Hartline; I promise I have nothing against him. He just happened to finish exactly 100th in passing efficiency, so that got him chosen as the representative for the bottom of the college football quarterback pecking order.

His adjusted passing efficiency is 97.6, and his adjusted passer rating is 62.3. The former is higher than the efficiency for the NFL’s worst regular, Derek Anderson, but the latter is lower than the former Oregon State turnover machine’s rating. In other words, they are about even when it comes to performance relative to their rankings within their respective leagues.

That is why there are a lot of college guys at the top of the hypothetical combined rankings. There would be a lot at the bottom of them too, and plenty in the middle as well. After all, there are 119 teams in Div. I-A but only 32 NFL teams.

Quarterbacking quality is a spectrum, and college football simply has more guys to put on its range than the NFL does.

My goal wasn’t to try to tell you that college quarterbacks are better than pro quarterbacks because, as I said at the beginning, that’s patently untrue. I only wanted to show how the two major systems of rating quarterbacks compare so you can have some sort of reference when seeing one or the other.

Neither method is perfect, and there might even be a better one out there. Until you can convince the NFL or NCAA to adopt it though, passer rating and passing efficiency the big ones we’ve all got.

Now, at least, you can eyeball the differences in them and make a pretty good guess as to how college and pro quarterbacks are doing relative to each other and their respective leagues.

2008 Review: Risers and Fallers

Before last season, I looked at teams’ records in close games to see if they were potential risers or fallers from the previous season. The rationale is that in close games, luck determines the outcome as much as anything. If a team was particularly lucky or unlucky in 2007, you’d figure they would regress to the mean in 2008 and rise or fall accordingly (all else being equal).

All else is not equal, so it’s an imperfect method. However few teams change dramatically from one season to the next, so it works pretty well at predicting which way the teams it singles out will go.

I chose to look at games decided by eight points or less because a touchdown and two-point conversion could tie them. If a team’s difference between close games won and lost was three or more, I labeled it a potential riser or faller. If the difference was two, I put the team on a “wait list.”

I later found out that this type of study is something Phil Steele puts in his magazine every season, but he uses seven points instead of eight. I also looked only at BCS teams, whereas I think he does all of I-A. Either way, here’s how the 2008 bunch (by my figures) made out.

Potential Risers

First, let’s start with the positive.

Seven teams were picked as risers: Maryland, Minnesota, Michigan State, UCLA, North Carolina, Vanderbilt, and Washington. In my writeup, I expressed doubt that UCLA was going to do it, and I hinted that Washington might not either.

As it turns out, those two Pac-10 teams were the only potential risers who did not improve their records. All others won at least two more games, and Minnesota even won a robust six more games than in 2007. Together, the potential risers improved their records by an average of 1.43 wins.

The teams that were wait listed were Alabama, Arizona, Cincinnati, Louisville, and Ole Miss. UL was the only underachiever of the bunch, and like Minnesota, Ole Miss saw a six-game uptick in its wins. Alabama was close, winning five more in 2008 than in 2007. Together, the wait list teams increased their win counts by an average of 2.8 games.

In all, the potential risers and wait list teams increased their win counts by an average of exactly two wins apiece.

Potential Fallers

Now on to the harbingers of doom.

Eight teams were identified as potential fallers: Arizona State, Boston College, Kansas, Kentucky, Mississippi State, Northwestern, Oregon State, and Virginia. I expressed some skepticism towards ASU and Northwestern falling off, and I ended up half right.

Northwestern was the only team of the bunch to improve its record, winning three additional games in 2008. Oregon State held steady with a 9-4 mark in two straight seasons. The Sun Devils fell the hardest, winning five fewer games in 2008. Together, the potential fallers won 2.13 fewer games than in 2007.

The teams on the wait list were UConn, LSU, NC State, Texas, and Wisconsin. The Wolfpack and Longhorns bucked the trend, winning one and two more games, respectively, in 2008 than they did in 2007. Defending champs LSU dropped off the farthest, winning four fewer games. Together, the wait list teams won 0.8 fewer games in 2008 than in 2007.

In all, the potential fallers and their wait list brethren won 1.62 fewer games than in 2007.

Conclusion

This particular form of prognostication was not 100 percent accurate, but I challenge you to find one that is. Eighteen of the 25 teams identified by this method went in the direction predicted and one held steady. Even counting Oregon State as a miss, that is still a 72 percent hit rate.

Who are the potential risers and fallers for 2009? Stay tuned because I haven’t run the numbers yet, but once I have them done I’ll let you know.